Accuracy of the acoustic Doppler velocimeter (ADV) is evaluated in this paper. Simultaneous measurements of open-channel flow were undertaken in a 17-m flume using an ADV and a laser Doppler velocimeter. Flow velocity records obtained by both instruments are used for estimating the true (“ground truth”) flow characteristics and the noise variances encountered during the experimental runs. The measured values are compared with estimates of the true flow characteristics and values of variance (〈u′2〉, 〈w′2〉) and covariance (〈u′w′〉) predicted by semiempirical models for open-channel flow. The analysis showed that the ADV sensor can measure mean velocity and Reynolds stress within 1% of the estimated true value. Mean velocities can be obtained at distances less than 1 cm from the boundary, whereas Reynolds stress values obtained at elevations greater than 3 cm above the bottom exhibit a variation that is in agreement with the predictions of the semiempirical models. Closer to the boundary, the measured Reynolds stresses deviate from those predicted by the model, probably due to the size of the ADV sample volume. Turbulence spectra computed using the ADV records agree with theoretical spectra after corrections are applied for the spatial averaging due to the size of the sample volume and a noise floor. The noise variance in ADV velocity records consists of two terms. One is related to the electronic circuitry of the sensor and its ability to resolve phase differences, whereas the second is flow related. The latter noise component dominates at rapid flows. The error in flow measurements due to the former noise term depends on sensor velocity range setting and ranges from ±0.95 to ±3.0 mm s−1. Noise due to shear within the sample volume and to Doppler broadening is primarily a function of the turbulence dissipation parameter. Noise variances calculated using spectral analysis and the results of the ground truthing technique are compared with theoretical estimates of noise.